Representing the behaviour of software projects using multi-dimensional timelines

Abstract

Context: There are few empirical studies in the empirical software engineering research community that describe software projects, at the level of the project, as they progress over time. Objective: To investigate how to coherently represent a large volume of qualitative and quantitative data on a range of project-level attributes as those attributes change over time. Method: Develop a modelling technique, multi-dimensional timelines (MDTs) and undertake a preliminary appraisal of the technique using examples from a longitudinal case study of a project at IBM Hursley Park. Results: MDTs can represent project-level attributes as they change over time, provided these attributes, and the empirical data about them, can be located in time (an analytical requirement) and can be represented in terms of the simple geometrical structures of points, lines and planes (a graphical requirement). Changes in attributes are documented at the point in time at which the change occurs. There are a number of ways in which an attribute can be represented on the MDT: as a quantitative time series, as an event, as an event with label containing complex qualitative information, or as a schedule. The MDT technique is currently not capable of representing relationships between different attributes e.g. a causal relationship. Conclusion: The initial appraisal of MDTs is encouraging, but further work is needed on the development of the MDT technique and on its evaluation.